Um...you should just stop talking.
:thumbsup:
Dredd is quite right about the Samsung MV-3V4G3D/US. It trumps my G.SKILL 8GBRL kit, except for its "recommended" or "certified" latency settings, but probably will run at those same timings at the rated speed with only a slight voltage bump. The GBRLs are rated at 1.50V -- and it's likely I will get them to run as two kits in four slots at about that voltage @ DDR3-1866. A 20-hour test the other day at that speed using HCI Memtest 4.0 -- 16GB 2x8GB -- got past 380% coverage using the bootable-CD version before forcing a system reboot.
I should only have to increase the VCCIO another 0.025V to 1.10V to get the G.SKILLs to pass considerably more than that, and if that doesn't work, 1.15V probably will.
In this context, the Samsungs have a lot going for them -- in fact, a lot more. Given the relationship between vDIMM and VCCIO, a low voltage like 1.35V in the Samsung spec means that the VCCIO setting might be lower than 1.0V. More certain -- if you only have to raise the vDIMM to 1.475V or just under 1.5, the VCCIO can vary over a wider range to get higher speeds than 1600, while low enough to be "safe" for the processor and its IMC.
I could almost spend the $96 to get me into a two-kit Samsung ball-game, but it looks like I'm going to get these G.SKILLs to work at settings I've chosen, and I don't need to take them to 2133 Mhz.
As for heatspreaders, G.SKILL has been a stand-up brand for performance, but they've never been above giving their product different flavors of "bling-bling." They use Samsung parts, or so I heard. But there may be some truth to the idea that the heatspreaders could be the most distinct difference for differently-spec'd RAM kits, howsoever a deliberate and painstaking binning process assures product quality for the rated speeds and timings.